Vol. 174, No. 19JOURNAL OF BACrERIOLOGY, OCt. 1992, p. 6011-60170021-9193/92/196011-07$02.00/0Copyright © 1992, American Society for Microbiology

A Conjugation Procedure for Bdellovibrio bacteriovorus andIts Use To Identify DNA Sequences That Enhance the

Plaque-Forming Ability of a SpontaneousHost-Independent Mutant

TODD W. COT11ER't AND MICHAEL F. THOMASHOWl 2*Department of Microbiology' and Department of Crop and Soil Sciences,2*

Michigan State University, East Lansing, Michigan 48824

Received 16 June 1992/Accepted 23 July 1992

Wild-type bdeliovibrios are obligate intraperiplasmic parasites of other gram-negative bacteria. However,spontaneous mutants that can be cultured in the absence of host cells occur at a frequency of 10-6 to 10-7. Suchhost-independent (H-I) mutants generally display diminished intraperiplasmic-growth capabilities and formplaques that are smaller and more turbid than those formed by wild-type strains on lawns of host cells. Ananalysis of the gene(s) responsible for the H-I phenotype should provide significant insight into the nature ofBdellovibrio host dependence. Toward this end, a conjugation procedure to transfer both IncQ and IncP vectorsfrom Escherichia coli to Bdellovibrio bacteriovorus was developed. It was found that IncQ-type plasmids werecapable of autonomous replication in B. bacteriovorus, while IncP derivatives were not. However, IncPplasmids could be maintained in B. bacteriovorus via homologous recombination through cloned B. bacterio-vorus DNA sequences. It was also found that genomic libraries of wild-type B. bacteriovorus 109J DNAconstructed in the IncP cosmid pVK100 were stably maintained in E. coli; those constructed in the IncQ cosmidpBM33 were unstable. Finally, we used the conjugation procedure and the B. bacteriovorus libraries to identifya 5.6-kb BamHI fragment of wild-type B. bacteriovorus DNA that significantly enhanced the plaque-formingability of an H-I mutant, B. bacteyiovorus BB5.

Members of the genus Bdellovibrio are obligate intra-periplasmic (IP) parasites of other gram-negative bacteria(11, 30, 31). Their unique development cycle consists of twofundamental stages, a free-swimming attack phase and an IPgrowth phase. While in the attack phase, bdellovibrios arehighly motile and metabolically active but they do notreplicate their DNA. Upon contact with a host, a bdellovi-brio attaches to the outer envelope of the cell and after ashort period rapidly penetrates into the host, becominglodged within the periplasmic space. During the invasionprocess, the bdellovibrio drops its flagellum and initiates thetransition from attack phase to growth phase. Early in thegrowth phase, the bdellovibrio converts the host cell into astable spherical structure, termed the bdelloplast, and de-grades host cell macromolecules to products that are usedfor energy generation and cell biosynthesis. DNA replicationis initiated about 30 min after invasion. During the remainingperiod of growth, the bdellovibrio elongates into a coiled,multicellular filament, which, at the cessation of growth,fragments into individual attack-phase cells. The residualbdelloplast is then lysed, and the progeny bdellovibrios arereleased into the environment.The transition of a wild-type bdellovibrio from an attack-

phase cell to a growth-phase cell is strictly dependent on theavailability of a suitable host; all attempts to bypass the hostcell requirement with commercial media have been unsuc-cessful (10, 13, 20, 26). Determining the molecular basis for

* Corresponding author. Electronic mail address: 22676mft@msu.edu (Bitnet).

t Present address: Laboratory of Intracellular Parasites, RockyMountain Laboratories, National Institute of Allergy and InfectiousDiseases, Hamilton, MT 59840.

this host-dependent (H-D) phenotype is critical to an under-standing of Bdellovibrio growth and development. Atpresent, however, little is known about this aspect of Bdell-ovibrio biology. Several studies have demonstrated thatconcentrated cellular extracts from hosts and other bacteriacan induce wild-type bdellovibrios to enter the growth phaseand support the completion of the entire growth cycle (10,13, 20). Further, Gray and Ruby (11) have hypothesized thatat least two different host signals are required to successfullycarry out the IP growth cycle: one to trigger differentiation ofan attack-phase cell into a growth-phase cell and a second toinitiate rounds of DNA synthesis. However, no specificgrowth factors or signal molecules have been identified, andthe mechanism by which the extracts stimulate Bdellovibriogrowth and multiplication remains unknown.An important discovery made early in the study of bdel-

lovibrios was that spontaneous mutants that no longer re-quire host cells for growth can be isolated (6, 14, 25, 27, 32).Such host-independent (H-I) mutants complete the transitionfrom attack phase to growth phase and back again onstandard complex bacteriological media without any factorsthat are specifically associated with the IP niche. Uponinitial isolation, the majority of these mutants retain limitedIP growth capabilities and are termed facultative. Whenplated on lawns of host cells, facultative H-I mutants formplaques that are smaller and more turbid than plaquesformed by wild-type bdellovibrios (6, 25, 32).H-I mutants have been reported to arise at a frequency of

10-6 to 10-7 (25, 32), suggesting that single mutationalevents at one or more loci can obviate the bdellovibriorequirement for host cells. The isolation and characteriza-tion of genes affected in H-I mutants should provide signif-icant insight into the nature ofBdellovibrio host dependence.

6011

on April 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

6012 COTTEIR AND THOMASHOW

TABLE 1. Bacterial strains and plasmids

Source orStrain or plasmid Description reference

B. bactenovorus109J109J.1109J.2BB5

E. coliML35SR-1DH5

SM1o

pKC7pBR328pRK2013pSUP204pSUP304.1pMMB33pRK290pVK100pVK102pVKa-1

pTC3

pTC5

pTC6

pTC7

pTC12

pTC50

Wild-typeSmr derivative of 109JRf' derivative of 109J.1H-I derivative of 109J.2

B lacI lacYSmr Rf' derivative of ML35F- endAl recAl hsdRl7(rK MK )deoR thi-1 supE44 gyrA96 relA1

supE44 hsdR thi-1 thr-1 leuB6lacYl tonA21 recA Muc+RP4-2Tc::Mu; Kmr

ColEl Apr KimrColEl Apr Cmr TcrColEl KMr tra(RK2)IncQ Apr Cmr TcrIncQ Apr KmrIncQ Kmr cos

IncP TcrIncP Tcr KMrIncP Tcr KmrDerivative of pVK100 containing

445-bp HaeII fragment frompUC19

23.5-kb insert of B. bacteriovorus109J DNA in EcoRI site ofpVK100

4.9-kb insert of B. bacteriovorus109J DNA in EcoRI site ofpVK100

20.5-kb insert of B. bacteriovorus109J DNA in EcoRI site ofpVK100

19.5-kb insert of B. bacteriovorus109J DNA in EcoRI site ofpVK100

5.6-kb BamHI fragment frompTC7 in BglIl site of pVK102

0.96-kb EcoRI-XbaI fragment frompTC12 in BamHl site of pVKa-1

This approach, however, has not been exploited because ofa lack of systems for genetically manipulating bdellovibrios.Here, we begin to rectify this situation. We identify vectorsthat can be used to clone Bdellovibrio bacteriovorus DNAand describe a conjugation procedure to introduce DNA intoboth wild-type and H-I strains of B. bacteriovorus. We thendescribe the use of the vectors and conjugation procedure toidentify DNA sequences from wild-type B. bacteriovorusthat enhance the plaque-forming ability of an H-I mutant.

(Portions of this work were presented previously in pre-liminary form [3, 4].)

MATERIALS AND METHODSBacterial strains and plasmids. The bacterial strains and

plasmids used are listed in Table 1. B. bacteriovorus 109Jand Escherichia coli ML35 were obtained from S. C. Ritten-berg. All Bdellovibrio strains were single-plaque or single-colony purified and stored in 15% glycerol at -80°C.Media and growth conditions. All E. coli cultures were

grown at 37°C in Luria-Bertani medium (17). When appro-

priate, E. coli cultures contained the following antibiotics atthe indicated concentrations (in micrograms per milliliter):ampicillin, 100; chloramphenicol, 20; kanamycin sulfate, 25;rifampin, 100; streptomycin sulfate, 50; and tetracycline, 12.

IP cultures of B. bacteriovorus 109J and its derivativeswere grown in dilute nutrient broth (DNB) at 30°C, with E.coli ML35 as the host cell. DNB consisted of 1 mM CaCl2,0.1 mM MgCl2, and 0.8 g of nutrient broth (Difco) per liter.Host cells were prepared by washing overnight cultures of E.coli ML35 once in an equal volume of DNB. Liquid IPcultures were established by adding approximately 109 hostcells and 107 bdellovibrios per ml, with single plaques orovernight cultures as the bdellovibrio inoculum. These cul-tures routinely lysed completely within 24 h. IP cultureswere plated for plaque development by adding 0.1 ml of theappropriate bdellovibrio dilution and 1010 washed host cells(in 0.3 ml) to 3 ml of an agar overlay (DNB plus 0.7% agarheld at 50°C) and immediately spreading the mixture onDNB plates that contained 1.5% agar. Under these condi-tions, plaques became visible after a 3- to 4-day incubation at30°C. Rf and Smr bdellovibrios were usually grown on E.coli SR-1 in the presence of 100 ,ug of rifampin per ml or 50,ug of streptomycin sulfate per ml. Plasmid-containing bdel-lovibrios were grown on E. coli ML35 carrying pKC7 orpBR328, in the presence of 35 ,ug of kanamycin sulfate perml or 10 ,ug of chloramphenicol per ml, respectively.

H-I bdellovibrio cultures were grown at 30°C in PYEmedium (25) that contained 10 g of peptone and 3 g of yeastextract per liter. PYE plates were solidified with 1.5% agar.When appropriate, antibiotics were used at the same con-centrations as in IP cultures.

Isolation and characterization of an H-I mutant. Spontane-ous H-I mutants of B. bacteiiovorus 109J were obtained at afrequency of 10-6 to 10-7 by the method of Seidler and Starr(25). Selection for H-I growth yielded yellow CFU thatvaried in size from barely perceptible to about 3 mm indiameter after a 7-day incubation at 30°C. The spontaneousH-I mutants that formed large colonies could generally besubcultured on solid or liquid PYE medium, whereas thosethat formed small colonies could not. A well-isolated large-colony mutant, B. bacteriovorus BB5, was selected forfurther characterization. This mutant formed circular,smooth-edged colonies that were about 2 mm in diameterafter 7 days of incubation on PYE plates at 30°C. B.bacteriovorus BB5 also formed plaques when plated inoverlay lawns of host cells on DNB medium, but the plaqueswere much smaller and more turbid than those formed bywild-type B. bacteriovorus 109J (see Results). The total PFUformed by B. bacteriovorus BB5 were generally 10 to 100%of the total number of CFU.

Chemicals and reagents. Complex medium componentswere purchased from Difco. Restriction endonucleases andT4 DNA ligase were purchased from New England Biolabs.[a-32P]dCTP (800 Ci/mmol) was purchased from DuPont/New England Nuclear.

Matings. Individual matings were conducted on 3-cm-square pieces of nitrocellulose (Schleicher and Schuell) thatwere incubated on PYE plates. The nitrocellulose wasautoclaved in water, placed on PYE plates, and allowed todry (30 min at room temperature [RT]). Wild-type B. bacte-novorus recipients were prepared from freshly lysed IPcultures. Such cultures were concentrated 10-fold by cen-trifugation, and 0.1 ml of the suspension was spread on anitrocellulose filter and allowed to dry (30 min at RT). H-Imutant recipients were prepared by placing 0.1 ml of anovernight culture on a nitrocellulose filter, letting it dry (30

J. BACT1ERIOL.

on April 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

BDELLOVIBRIO CONJUGATION PROCEDURE 6013

min at RT), and then incubating the filter overnight at 30°Con a PYE plate. Donor E. coli cultures that had been washedonce in DNB and concentrated 10-fold were spread (0.1 ml)on top of the recipients. After 16 to 24 h of incubation at30°C, individual matings (nitrocellulose filters) were trans-ferred to 2 ml of DNB and vortexed vigorously and thenserially diluted and plated for PFU and CFU. When cultureswere plated for axenic growth on PYE plates, streptomycinsulfate (50 ,ug/ml) was included in the medium to selectagainst growth of the donor. All bdellovibrio recipient cul-tures were started from -80°C stocks. Donor strains wereeither E. coli SM10 or E. coli DHS. For E. coli SM10,functions required for the conjugal transfer of IncQ- andIncP-type plasmids were provided by an IncP plasmid that isintegrated into the SM10 genome (28). When E. coli DH5was used as the donor, the same transfer functions wereprovided by the helper plasmid pRK2013 (8), a ColElderivative that cannot replicate in B. bacteriovorus (2a).When E. coli DH5 was the donor, overnight cultures of thestrain containing the target plasmid and E. coliDHS(pRK2013) were mixed in equal volumes and used asthe donor suspension, as described above.DNA manipulations, Southern analysis, and library con-

struction. Most DNA purification and recombinant DNAmethods were standard (23). B. bacteniovorus genomic DNAwas purified by a CTAB (cetyltrimethylammonium bro-mide)-based extraction procedure (1).For Southern analysis, B. bacteriovorus genomic DNA

was digested with various restriction enzymes, fractionatedby electrophoresis in 0.7% agarose gels, and transferred toNytran membranes (Schleicher and Schuell) by the capillarymethod. Prior to hybridization, membranes were prewashedin a buffer containing 0.lx SSPE (lx SSPE is 0.18 M NaCl,10 mM NaPO4, and 1 mM EDTA [pH 7.7]) and 0.5% sodiumdodecyl sulfate (SDS) at 65°C. Prewashed membranes werethen prehybridized for 1 h at 68°C in hybridization fluid thatcontained 6x SSPE, 0.5% SDS, and 0.25% nonfat dry milk(Sanalac). A radiolabeled probe was then added and allowedto hybridize overnight at 68°C. All posthybridization washescontained 0.5% SDS and were done in the following order:twice in 2x SSPE at RT, twice in O.lx SSPE at RT, andthree times in O.1x SSPE at 68°C. Radiolabeled probes(approximately 108 dpm/,Lg of DNA) were produced by nicktranslation (kit obtained from Bethesda Research Laborato-ries) or random priming (23).

B. bacteriovorus 109J genomic libraries were constructedby the method described by Ausubel et al. (1). GenomicDNA was partially digested with EcoRI and size fractionatedin 0.5% agarose gels, and the DNA fragments ranging in sizefrom 20 to 30 kb were electroeluted from the gel and purifiedwith Elutip-d columns (Schleicher and Schuell). The size-fractionated DNA was ligated into the EcoRI site of pVK100(15), and the products were packaged into lambda particlesby using commercial extracts (Promega) according to sup-plier specifications. Packaged cosmids were transduced intoE. coli DH5 and stored at -80°C.

RESULTS

Conjugal transfer of RSF1010 and RK2 derivatives into B.bacteriovorus. It has been shown previously that RK2 trans-fer functions supplied in trans can effect conjugal transfer ofRSF1010 (IncQ)- and RK2 (IncP)-derived plasmids betweengram-negative bacteria (7, 18). We therefore attempted touse RK2 transfer functions to conjugally transfer bothclasses of plasmids from E. coli to B. bacteriovorus (see

TABLE 2. Conjugal transfer of plasmids into B. bactenovorusa

Plasmid Transfer Useful antibiotic(s) forfrequencyb selection

IncQpSUP204 10-3 Chloramphenicol, kanamycinpSUP304.1 10-3 KanamycinpMMB33 10-3 Kanamycin

IncPpRK290 NDCpVK100 NDpTC3 10-4 Kanamycina Data apply to matings involving both H-D and H-I strains as recipients,

with both SM10 and DH5(pRK2013) as donors.b Expressed as approximate number of antibiotic-resistant recipients per

total number of recipients. Frequencies given are typical of those obtained innumerous experiments.

c ND, not detected (less than 10-').

Materials and Methods). Matings of E. coli carryingRSF1010 derivative pSUP204, pSUP304.1, or pMMB33 withwild-type B. bacteriovorus 109J.2 and H-I B. bactenovorusBBS yielded antibiotic-resistant recipients (Table 2). Kana-mycin sulfate (20 to 40 ,ug/ml) and chloramphenicol (5 to 10,ug/ml) were effective in selecting for transfer of the RSF1010derivatives, while tetracycline (2 to 25 ,ug/ml) and ampicillin(5 to 50 ,ug/ml) were not. Matings conducted in the absenceof RK2 transfer functions did not yield antibiotic-resistantbdellovibrios, indicating that plasmid transfer was conjugalin nature. Southern analysis indicated that the RSF1010derivatives were maintained in the B. bacteriovorus strainsby autonomous replication: BamHI digestion of total DNAisolated from B. bacteriovorus BB5(pMMB33) produced asingle 13.8-kb band, the size expected for linear pMMB33,that hybridized with pMMB33 (Fig. 1A). In addition, trans-formation of E. coli DH5 with total DNA from B. bactenio-vorus BB5(pMMB33) yielded transformants that containedpMMB33 (data not shown).

In contrast to results with the RSF1010-derived plasmids,

A1 2

kbp

_ -13.8

B2

kbp

* -5.6

OM -4.1

-2.4

FIG. 1. Southern analysis of plasmid-containing B. bacterio-vorus strains. (A) BamHI digests of total DNA isolated from B.bacteriovorus BB5 (lane 1) and BB5(pMMB33) (lane 2) hybridizedwith radiolabeled pMMB33. (B) BamHI digests of total DNAisolated from B. bacteniovorus BB5 (lane 1) and BB5(pTC50) (lane 2)hybridized with a radiolabeled 5.6-kb BamHI insert from pTC12.

VOL. 174, 1992

on April 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

6014 COlTER AND THOMASHOW

FIG. 2. Plaque phenotypes of B. bacteriovorus strains. Theplaques formed by B. bacteriovorus 109J.2(pTC3) (A), BB5(pTC3)(B), and BBS(pTC12) (C) are shown. pTC3 is used here as a controlto confer Kmr; it does not affect the plaque phenotype of wild-typeor mutant strains. Overnight IP cultures of each strain yielded about109 PFU/ml (5).

all attempts to conjugally transfer the RK2-based vectorspRK290 and pVK100 into B. bacteriovorus 109J and B.

bacteriovorus BB5 failed to yield antibiotic-resistant recipi-ents (Table 2). Since RSF1010 plasmids were mobilized intoB. bacteriovorus by RK2 transfer functions, it was possiblethat RK2 plasmids were also transferred but could notreplicate. If true, insertion of a B. bacteriovorus DNAsequence into an RK2 derivative could potentially allow thevector to be maintained in bdellovibrios via integration byhomologous recombination. Indeed, conjugal transfer ofpTC3, a derivative of pVK100 containing a random 23.5-kbfragment of B. bacteriovorus DNA (Table 1), producedkanamycin-resistant recipients of both wild-type and H-I B.

bacteriovorus (Table 2).Integration of an RK2-based construct into the B. bacte-

riovorus genome was demonstrated by Southern analysis oftotal DNA obtained from B. bacteriovorus BB5(pTC50) (Fig.1B). pTC50 contains a 0.96-kb EcoRI-XbaI fragment of B.bacteriovorus 109J DNA subcloned from the 5.6-kb BamHIfragment of B. bacteriovorus 109J DNA contained in pTC12(Table 1). The XbaI and EcoRI termini of the 0.96-kbEcoRI-XbaI fragment, however, were replaced with BamHItermini prior to cloning into the BamHI site of the RK2-based vector pVKa-1 (5). On the basis of the DNA sequenceof the 0.96-kb EcoRI-XbaI fragment (5), integration ofpTC50 into the B. bacteriovorus BB5 genome via homolo-gous recombination should have transformed the 5.6-kbBamHI fragment into two BamHI fragments of 4.1 and 2.4kb. This prediction proved to be the case: total DNA from B.bacteriovorus BB5 contained a single hybridizing band rep-resenting the 5.6-kb BamHI fragment, while a BamHI digestof B. bacteriovorus BB5(pTC50) contained two smallerfragments of 4.1 and 2.4 kb (Fig. 1B).

Identification of wild-type B. bacteriovorus DNA sequencesthat enhance plaque formation by H-I mutant B. bacteriovorusBBS. The H-I mutant B. bacteriovorus BB5, like previouslydescribed facultative H-I mutants (6, 24), retains a limitedcapacity for IP growth and forms small, turbid plaques onlawns of host cells (Fig. 2B). If this phenotype was due to amutation that resulted in gene inactivation, then it might bepossible to identify the affected gene by transferring awild-type genomic DNA library into B. bactenovorus BB5and screening for recombinants with an enhanced plaquingability (i.e., recombinants that produce large, clear plaques).The wild-type sequences that had been transferred to theenhanced-plaque recombinants could then be isolated andcharacterized.To conduct the experiment described above, we first

needed to construct a genomic library of B. bacteriovorus.

J. BAcTERiOL.

on April 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

BDELLOVIBRIO CONJUGATION PROCEDURE 6015

This was initially attempted with the RSF1010-based cosmidpMMB33, but it was found that this vector could not stablymaintain large B. bacteriovorus DNA inserts (25 to 35 kbp)in E. coli. Stable libraries of B. bactenovorus genomic DNA,however, could be constructed with the RK2-based cosmidpVK100. Six independently packaged libraries, VL-1through VL-6, used in subsequent experiments were com-posed of about 100 clones each. Given an insert size ofbetween 22 and 28 kbp and a B. bacteriovorus genome sizeof 1 x 106 to 2 x 10' bp (16, 24), it was calculated by themethod of Clarke and Carbon (2) that about 360 cloneswould be required to ensure a 99% probability that any givensequence would be represented in a library. The six librarieswere then individually mated into B. bacteriovorus BB5, andthe recombinants were screened for enhanced plaquing. Noenhanced-plaque recombinants were obtained with librariesVL-3 and VL-4 or plasmid pTC3 (which served as thenegative control). However, about 1% of the total Kmrrecipients obtained with libraries VL-1, VL-2, VL-5, andVL-6 gave rise to plaques that were much larger and clearerthan those produced by B. bacteriovorus BB5.

If the observed enhancement of B. bacteriovorus BB5plaquing activity resulted from homologous recombinationof wild-type sequences into the recipient genome at the siteof the H-I mutation, then identical or related cosmids shouldhave been present in the enhanced-plaque recombinants.Southern analysis was conducted to determine whether thiswas the case. Total DNA from a number of recombinantswas digested with HindIII, an enzyme that cuts once withinpVK100, and was subjected to Southern analysis withpVK100 as the probe (Fig. 3). Each digest would be ex-pected to contain two bands that hybridized to the cloningvector, both of which would have extended in oppositedirections from the HindIII site within pVK100 to HindIIIsites in the adjacent bdellovibrio DNA (Fig. 3A). When 18randomly picked recipients from the VL-1 and VL-2 matingswere analyzed, 17 distinct hybridization patterns were ob-served (Fig. 3B). These results contrasted with those ob-tained with enhanced-plaque recombinants obtained fromthe VL-1, VL-2, and VL-5 matings. All of the B. bactenio-vorus BB5 recombinants that contained cosmids from VL-1(nine individuals) and VL-2 (eight individuals) displayed thesame hybridization pattern, and a second pattern was seen in10 isolates that contained cosmids from VL-5 (Fig. 3C; thehybridization patterns of six individuals from each matingare presented). Thus, the data indicated that specific regionsof the B. bacteriovorus 109J genome were involved inconferring the enhanced-plaque phenotype upon B. bacteri-ovorus BB5.

Further analysis of the wild-type DNA sequences respon-sible for plaque enhancement required their isolation. Thiswas accomplished in several steps. Total DNA from anenhanced-plaque recombinant that showed the predominantrestriction fragment hybridization pattern (Fig. 3C, VL-1lane 5) was digested with BamHI, thereby releasing thevector from the genome with flanking bdellovibrio sequencesattached to each end. This linear, vector-containing BamHIfragment was then circularized in a dilute ligation andtransformed into E. coli DH5. A single plasmid, pTC5, thatcontained pVK100 plus two flanking EcoRI-BamHI frag-ments of 2.3 and 2.6 kb was identified. The 4.9-kb EcoRIinsert from pTC5 was then purified and used to probe colonylifts of the VL-1 library. Two cosmids that hybridized to thepTC5 insert, pTC6 and pTC7, were isolated and found tocontain overlapping inserts of 20.5 and 19.5 kb, respectively(Fig. 4).

A

H

H Q

x

H H

I I I I

BBVL-1 VL-2

kbp23 - 0 *n~~~~~~~~~~

411w I,

14m

3.7-

cVL-1 VL-2 VL-5

kbp

9.4-

4.4

2.3-

FIG. 3. Southern analysis of cosmid-containing recombinants.(A) Schematic diagram showing cointegration of a cosmid into therecipient B. bacteriovorus genome. Thin line, recipient genome;solid box, cloned B. bacteriovorus DNA; cross-hatched box,pVK100 vector. Two hypothetical HindIII fragments that hybridizeto the pVK100 probe are shown. H, HindlIl restriction site. (B)Southern analysis of HindIII digests of total DNA isolated from 18random B. bacteriovorus BB5 recipients hybridized with radiola-beled pVK100. Individual Kmr isolates were obtained after matingwith libraries VL-1 and VL-2. (C) Southern analysis of HindIlldigests of total DNA isolated from 18 enhanced-plaque B. bacteri-ovorus BB5 recombinants hybridized with radiolabeled pVK100.Individual Kmr isolates were obtained after mating with librariesVL-1, VL-2, and VL-5.

Recombination of pTC6 into the B. bacteriovorus BB5genome had no effect on the plaquing phenotype of the H-Imutant. However, recombination of pTC7 into B. bacterio-vorus BB5 did; it conferred the enhanced-plaque phenotype.The region of pTC7 responsible for the phenotype was thenfurther delineated. In particular, the 5.6-kb BamHI fragmentof pTC7 was subcloned, yielding pTC12 (Fig. 4), and wasfound to enhance the plaque-forming ability of B. bacterio-vorus BB5 (Fig. 2C). The plaques produced by B. bacterio-vorus BB5 appeared identical to those produced by B.bacteriovorus BB5(pTC7) (not shown) and nearly identicalto those formed by wild-type B. bacteriovorus (Fig. 2A),except possibly for a slight reduction in size.

DISCUSSIONMost of what has been learned about Bdellovibrio growth

and development has come from biochemical, physiological,

H

0---

VOL. 174, 1992

8.5-

5.6-ow

on April 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

6016 COTIER AND THOMASHOW

PlaqueEnhancement

pTC6

pTC7

pTC12

E E E E EI I i I

E BE E B E EBB EI 1 I I I I

FIG. 4. Abilities of wild-type B. bacteriovorus 109J DNA se-

quences to enhance plaque formation of B. bacteriovorus BBS. TheDNA inserts in pTC6, pTC7, and pTC12 are shown, and the abilitiesof the inserts to enhance plaque formation of B. bacteriovorus BB5are indicated. Restriction sites: B, BamHI; E, EcoRI.

and observational studies (11, 31). Mutational analysis hasalso been conducted, but the approach has been severelylimited by a lack of methods for genetically manipulatingbdellovibrios. Here we begin to address this deficiency. Inparticular, we have shown that IncQ and IncP plasmids canbe conjugally transferred from E. coli to both wild-type andH-I mutants of B. bacteriovorus. The IncQ plasmids aremaintained in B. bacteriovorus by autonomous replication.IncP plasmids cannot autonomously replicate in B. bacteri-ovorus but can be maintained via homologous recombinationand integration through cloned B. bacteriovorus sequences.We have also found that cosmid libraries of B. bacteriovorusDNA constructed with pVK100, an IncP vector, can bestably maintained in E. coli and can be transferred to B.bacteriovorus by conjugation.We have exploited the ability to transfer DNA from E. coli

to B. bacteriovorus to initiate a genetic analysis of theBdellovibrio requirement for host cells. Specifically, we haveidentified a 5.6-kb BamHI fragment of wild-type B. bacteri-ovorus 109J DNA that dramatically enhances the plaque-forming capacity of an H-I mutant, B. bacteriovorus BB5;whereas B. bacteriovorus BB5 forms small, turbid plaqueson lawns of E. coli, the recombinant B. bacteriovorusBB5(pTC12) forms large, clear plaques that closely resemblethose formed by wild-type B. bacteriovorus 109J (Fig. 2).The basic question raised, of course, is whether B. bacten-ovorus BB5 contains a mutation within the 5.6-kb BamHIfragment: i.e., was the enhanced-plaque phenotype of B.bacteriovorus BB5(pTC12) due to correction of the originalmutation or did recombination of the wild-type 5.6-kbBamHI fragment into the genome of B. bacteriovorus BB5indirectly suppress the poor-plaquing phenotype of the mu-tant? In an accompanying paper (5), we show that the formeris the case and define a genetic locus, designated hit (hostinteraction), that has a fundamental role in the Bdellovibrio-host cell interaction.

ACKNOWLEDGMENTS

This work was supported in part by the Michigan AgricultureExperiment Station and by Department of Energy contract grantDE-AC02-76ERO-1338. T.W.C. was a recipient of a Michigan StateUniversity Graduate Scholarship.

REFERENCES1. Ausubel, F. M., R. Brent, R. E. Kingston, D. D. Moore, J. G.

Seidmon, J. A. Smith, and K. Struhl. 1987. Current protocols inmolecular biology. Greene Publishing Associates, Wiley Inter-science, New York.

2. Clarke, L., and J. Carbon. 1976. A colony bank containingsynthetic ColEl hybrid plasmids representative of the entire E.coli genome. Cell 9:91-99.

2a.Cotter, T. Unpublished observation.3. Cotter, T., and M. Thomashow. 1991. Genetic analysis of

host-independent mutants of Bdellovibnio bacteriovorus 109J,abstr. I-17, p. 193. Abstr. 91st Gen. Meet. Am. Soc. Microbiol.1991.

4. Cotter, T. W., and M. F. Thomashow. 1989. Genetic systems forBdellovibrio bacteriovorus, abstr. I-70, p. 229. Abstr. 89thAnnu. Meet. Am. Soc. Microbiol. 1989.

5. Cotter, T. W., and M. F. Thomashow. 1992. Identification of aBdellovibnio bacteriovorus genetic locus, hit, associated withthe host-independent phenotype. J. Bacteriol. 174:6018-6024.

6. Diedrich, D. L., C. F. Denny, T. Hashimoto, and S. F. Conti.1970. Facultatively parasitic strain of Bdellovibrio bacterio-vorus. J. Bacteriol. 101:989-996.

7. Ditta, G., S. Stanfield, D. Corbin, and D. R. Helinski. 1980.Broad host range DNA cloning system for gram-negative bac-teria: construction of a gene bank of Rhizobium meliloti. Proc.Natl. Acad. Sci. USA 77:7347-7351.

8. Figurski, D., and D. Helinski. 1979. Replication of an origincontaining derivative of plasmid RK2 dependent on a plasmidfunction provided in trans. Proc. Natl. Acad. Sci. USA 76:1648-1652.

9. Frey, J., M. Bagdasarian, D. Feiss, F. C. H. Franklin, and J.Deshusses. 1983. Stable cosmid vectors that enable the introduc-tion of cloned fragments into a wide range of gram-negativebacteria. Gene 24:299-308.

10. Friedberg, D. 1978. Growth of host-dependent Bdellovibrio inhost cell free system. Arch. Microbiol. 116:185-190.

11. Gray, K. M., and E. G. Ruby. 1991. Intercellular signalling inthe Bdellovibrio developmental cycle, p. 333-366. In M. Dwor-kin (ed.), Microbial cell-cell interactions. American Society forMicrobiology, Washington, D.C.

12. Hanahan, D. 1985. Techniques for transformation of E. coli, p.109. In D. M. Glover (ed.), DNA cloning: a practical approach,vol. 1. IRL Press, McLean, Va.

13. Horowitz, A. T., M. Kessel, and M. Shilo. 1974. Growth cycle ofpredacious bdellovibrios in a host-free extract system and someproperties of the host extract. J. Bacteriol. 117:270-282.

14. Ishiguro, E. E. 1973. A growth initiation factor for host-inde-pendent derivatives of Bdellovibrio bacteriovorus. J. Bacteriol.115:243-252.

15. Knauf, V. C., and E. W. Nester. 1982. Wide host range cloningvectors: a cosmid clone bank of an Agrobacterium Ti plasmid.Plasmid 8:45-54.

16. Matin, A., and S. C. Rittenberg. 1972. Kinetics of deoxyribonu-cleic acid destruction and synthesis during growth of Bdellov-ibrio bacteriovorus strain 109D on Pseudomonas putida andEscherichia coli. J. Bacteriol. 111:664-673.

17. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

18. Priefer, U. B., R. Simon, and A. Puihler. 1985. Extension of thehost range of Escherichia coli vectors by incorporation ofRSF1010 replication and mobilization functions. J. Bacteriol.163:324-330.

19. Rao, R. N., and S. G. Rogers. 1979. Plasmid pKC7: a vectorcontaining ten restriction endonuclease sites suitable for cloningDNA segments. Gene 3:79-85.

20. Reiner, A. M., and M. Shilo. 1969. Host-independent growth ofBdellovibrio bacteriovonrs in microbial extracts. J. Gen. Micro-biol. 59:401-410.

21. Rittenberg, S. C. 1972. Nonidentity of Bdellovibrio bacterio-vorus strains 109D and 109J. J. Bacteriol. 109:432-433.

22. Rittenberg, S. C., and M. Shilo. 1970. Early host damage in theinfection cycle of Bdellovibrio bacteriovorus. J. Bacteriol. 102:149-160.

23. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecularcloning: a laboratory manual, 2nd ed. Cold Spring HarborLaboratory, Cold Spring Harbor, N.Y.

24. Seidler, R. J., M. Mandel, and J. N. Baptist. 1972. Molecularheterogeneity of the bdellovibrios: evidence of two new species.

J. BACTERIOL.

on April 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

VOL. 174, 1992 BDELLOVIBRIO CONJUGATION PROCEDURE 6017

J. Bacteriol. 109:209-217.25. Seidler, R. J., and M. P. Starr. 1969. Isolation and characteri-

zation of host-independent bdellovibrios. J. Bacteriol. 100:769-785.

26. Shilo, M. 1973. Rapports entre Bdellovibrio et ses hotes. Naturede la dependence. Bull. Inst. Pasteur 71:21-31.

27. Shilo, M., and B. Bruff. 1965. Lysis of gram-negative bacteria byhost independent ectoparasitic BdOllovibrio bacteriovorus iso-lates. J. Gen. Microbiol. 40:317-328.

28. Simon, R., U. Priefer, and A. Pfihler. 1983. A broad host rangemobilization system for in vivo genetic engineering: transposonmutagenesis in gram-negative bacteria. Bio/Technology 1:784-790.

29. Soberon, X., L. Covarrubias, and F. Bolivar. 1980. Construction

and characterization of new cloning vehicles. IV. Deletionderivatives of pBR322 and pBR325. Gene 9:287-305.

30. Thomashow, M. F., and T. W. Cotter. 1992. Bdellovibrio hostdependence: the search for signal molecules and genes thatregulate the intraperiplasmic growth cycle. J. Bacteriol. 174:5767-5771.

31. Thomashow, M. F., and S. C. Rittenberg. 1979. The intraperi-plasmnic growth cycle-the life style of the bdellovibrios, p.115-138. In J. H. Parish (ed.), Developmental biology of pro-karyotes. University of California Press, Berkeley.

32. Varon, M., and J. SeUffers. 1975. Symbiosis-independent andsymbiosis-incompetent mutants of Bdellovibrio bacteniovorus109J. J. Bacteriol. 124:1191-1197.

on April 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from